ANFIS-based Parameter Estimation of a Single Phase Inverter Circuit with Isolation Transformer

Abstract

In this study, it is aimed to isolate the output interface of single-phase inverter circuits and the grid from each other. For this purpose, three-dimensional isolation transformers are designed with a software using the In order to determine the leakage inductance behavior of the isolation transformer, whose electromagnetic modeling was carried out with Ferrite N87 core material, the geometric distances of the primary-secondary windings and the diameter sizing of the windings were changed in parametric linear steps, and a data set was obtained as a specific parametric scenario for the leakage inductance values. Thus, the estimation process of the electromagnetic modelling of the transformer has been successfully carried out thanks to the training and testing processes of Adaptive Network Based Fuzzy Inference Systems (ANFIS) with the numerical information obtained from the finite element analysis (FEA) parametric data set. After the estimation processes, the percentage error is calculated as 0.3470% and 0.4448% for training and testing. Thus, the determination of the isolation transformer with the optimum values designed for the inverter circuit has become easier. Also, experimental analysis is performed on inverters to prove the robustness of the proposed method. In this context, first of all, RMS values that vary according to the different operating parameters of the inverters are created. The proposed ANFIS based system estimates RMS values with 7,057 % error.